Wind Turbine Noise Prediction Of Multi-source Data Fusion

With the energy and environmental crisis is becoming more and more serious.The global scope is attached great importance to the development of renewable energy.China as a large population and energy,the development of renewable energy is increasing.Taking into account the economic viability of wind power,China has invested a lot of manpower and resources on wind turbine research and development,and wind turbine research and development unit also toward large-scale,large capacity direction.However,with the increasing of wind turbine generator sets,the length of blades is also proportional to the growth,which also makes the noise level if wind turbines gradually increased.The research of domestic and foreign scholars shows that the main noise of wind turbine is aerodynamic noise,while the main noise in wind turbine is structural noise.The noise parameter of wind turbine is related to the surrounding environment.At the same time,the noise signal parameter is also an important index to evaluate the quality of the large-scale wind turbine,and contains a lot of useful information when the wind turbine operation.Therefore,it is very important to research and analyze the noise signals of wind turbines.In this paper,the aerodynamic noise signals radiated by wind turbines are studied firstly and the aerodynamic noise signals of wind turbines are mainly caused by fan blades.The combined with regression analysis,support vector machine regression based on genetic algorithm and data fusion theory,the aerodynamic noise signals of wind turbines are filtered and eliminated,then the fusion prediction is carried out.Secondly,this paper studies the structural noise signals of wind turbine generator system,and studies the correlation between vibration signal and noise signal of wind turbine.Then,the structure noise of wind turbine is predicted by GA-SVR and data fusion theory.My main research contents are:1?A wind farm in Xinjiang Dabancheng wind area,according to the national standard IEC 61400-11 measurement standard,our research group carried on the noise signal measurement to the GW87/1500 wind turbine generator set in the normal operation state,this provides useful data for prediction of aerodynamic noise signals of wind turbines.2?The research shows that the direct acquisition of noise signals to wind turbines is not only difficult to operate,but also influenced by environmental factors.Therefore,this paper proposes to predict the noise signal through the non-acoustic signals of the wind turbine which can be easily collected,and to carry out the variable selection of the non-acoustic signals by regression analysis,eliminating the variables and data points with stronger collinearity.Then,Hilbert Huang(HHT)and wavelet analysis were used to extract the non-acoustic signals.Finally,a suitable noise prediction model is designed by improved genetic algorithm based support vector machine regression(GA-SVR),and the model is trained to achieve the feature level fusion prediction of the noise signal.3?Using the SVAN957 and EMT690,a 20 kW permanent magnet synchronous wind turbine generator set in the wind energy technology laboratory was experimentally investigated.Experiments were carried to simulate the various operating states of the wind turbines,and the correlation between the vibration signals and the noise signals was analyzed.Then,Hilbert Huang(HHT)and wavelet analysis were used to extract to extract the vibration signal.Finally,an improved algorithm based support vector machine regression(GA-SVR)was used to predict the feature level of the noise signal.Experiments were carried out to simulate the various operating states of the wind turbines,and the correction between the vibration signals and the noise signals was analyzed.Then,Hilbert Huang(HHT)and wavelet analysis were used to extract the vibration signal.Finally,an improved genetic algorithm based support vector machine regression(GA-SVR)was used to predict the feature level of the noise signal.The prediction of noise signal of wind turbine based on data fusion is summarized,and some prospect analysis made.